[cfe-dev] [llvm-dev] [RfC] A proposal of adding SPIR-V Toolchain in Clang

Mon Oct 29 09:54:25 PDT 2018

Thanks to everyone who participated in the Round Table discussion during the LLVM Dev Meeting. Below is a brief summary.

First of all, a number of concerns were raised again regarding the backend implementation. The implementation effort of developing the backend is significant and it's not clear to the SPIR-V developers if it results in the comparable or better quality generated binaries compared to the current solution using the IR translation [1]. This has been discussed earlier on llvm-dev as well [2], [3]. Another big concern was about the code duplication between the translation from/to SPIR-V that would have to be separated out into different flows.

The opinion of the developers attending the meeting was that we would prefer to continue using the translation tool [1]; but we would still like to improve the integration with LLVM as much as possible. There are a number of options that we have discussed briefly: external translator tool invocation via fork of Clang, adding SPIR-V triple w/o any external tool, adding translator as an official tool to the LLVM project (suggested earlier [4]).

To address concerns and suggestions from the LLVM community on this thread I am planning to look at the following topics with some help from Neil Hickey (in CC), before getting back with an updated proposal:

A. We will prototype a SPIR-V backend using the Global ISel in order to understand further the development effort and the quality of the code produced by the backend (+ potentially compilation time) compared to the simple IR translation.

B. I will look at how external tools are currently used with LLVM and try to suggest reasonable flow/process for the translator tool [1] that we would like to integrate at least for the purpose of importing SPIR-V modules to LLVM [5].

Please, let me know if you would like to be involved with any of the investigation activities I am planning or if you have any additional information that should be considered.

Cheers,
Anastasia

[1] https://github.com/KhronosGroup/SPIRV-LLVM-Translator
[2] http://lists.llvm.org/pipermail/llvm-dev/2017-May/112651.html
[3] http://lists.llvm.org/pipermail/llvm-dev/2017-May/112550.html

[4] https://lists.llvm.org/pipermail/llvm-dev/2018-February/121317.html

<https://github.com/KhronosGroup/SPIRV-LLVM-Translator>[5] <https://github.com/KhronosGroup/SPIRV-LLVM-Translator/wiki/SPIRV-Toolchain-for-Clang> http://lists.llvm.org/pipermail/cfe-dev/2018-September/059368.html

________________________________
From: llvm-dev <llvm-dev-bounces at lists.llvm.org> on behalf of Anastasia Stulova via llvm-dev <llvm-dev at lists.llvm.org>
Sent: 17 September 2018 10:00
To: Nicholas Wilson
Cc: llvm-dev at lists.llvm.org; nd; clang developer list; Brian.Sumner at amd.com
Subject: Re: [llvm-dev] [RfC] A proposal of adding SPIR-V Toolchain in Clang

Hi Nicholas,

A number of developers that are interested in the SPIR-V support will be attending the conference. Sounds good to have a round table. It is important to involve relevant developers outside of Khronos that are affected too!

So far there are the following topics to address:

- Generation of SPIR-V binaries in LLVM.

- Importing and linking SPIR-V libraries/binaries.

- Generic target in Clang for the next OpenCL C and C++ features.

- Use of external tools.

My only concern now is about the OpenCL C++ implementation that might be stalled while we are waiting for the decisions. But it's probably important to make sure we are going the right route to fit all requirements.

Anastasia

________________________________
From: Nicholas Wilson <iamthewilsonator at hotmail.com>
Sent: 17 September 2018 04:42:32
To: Anastasia Stulova
Cc: clang developer list; llvm-dev at lists.llvm.org; nd; Brian.Sumner at amd.com
Subject: Re: [llvm-dev] [RfC] A proposal of adding SPIR-V Toolchain in Clang

Given the amount of discussion that has gone on (and in past threads) and the size of the proposal, should we do a roundtable at the dev meeting in October to make sure everyone is on the same page going forward? I would come along for that if there is sufficient participation from Khronos to make good progress.

Nic

> On 10 Sep 2018, at 11:10 pm, Anastasia Stulova via llvm-dev <llvm-dev at lists.llvm.org> wrote:
>
> Hello,
>
> Since 2015 Khronos has switched to the new portable intermediate format SPIR-V, which has replaced the original SPIR. The advantage is that it offers higher portability across different toolchains. There was a talk about it at a Dev Meeting:
> http://llvm.org/devmtg/2017-03//2017/02/20/accepted-sessions.html#17
>
> LLVM currently only supports SPIR format for OpenCL in Clang. Several Khronos vendors (ARM, AMD, Intel, Xilinx, Codeplay and others) are interested in adding support for SPIR-V, which should gradually replace the old SPIR once products are no longer shipped with the old format. Here is the detailed description:
> https://github.com/KhronosGroup/SPIRV-LLVM-Translator/wiki/SPIRV-Toolchain-for-Clang
>
> To summarize, the idea is to add a SPIR-V target triple to Clang that can be used to generate a SPIR-V binary for OpenCL code. There was a separate thread regarding generation of SPIR-V binary and the community suggested that a translator from LLVM IR to SPIR-V can be used as an external tool, called llvm-spirv. This can be invoked similar to such tools as ptxas and fatbinary for the CUDA toolchain:
> http://lists.llvm.org/pipermail/llvm-dev/2018-February/121440.html
>
> An example of how Clang can be used to target SPIR-V:
>
> clang -c test.cl -target spirv[32|64]-unknown-unknown -o test.spv
>
> This will result in the following Clang actions:
>
> (1) clang -cc1 -triple spirv[32|64]-unknown-unknown test.cl -emit-llvm-bc -o test.bc
>
> (2) llvm-spirv test.bc -o test.spv
>
> SPIR-V generation is essential for completion of OpenCL C++ support in Clang, as newer OpenCL standards require frontend invocation to be performed offline, producing the SPIR-V binary that can be then loaded at application execution time. Besides that, it will also allow Clang to be used as a complete standalone tool to generate portable binaries that can then be consumed by different proprietary toolchains. In addition, this will open a path to the LLVM backends for various languages and frontends that already generate SPIR-V.
>
> A more detailed explanation of the complete design proposal is given in this Wiki page:
> https://github.com/KhronosGroup/SPIRV-LLVM-Translator/wiki/SPIRV-Toolchain-for-Clang
>
> Looking forward to any feedback about the proposal or possible collaborations,
>
> Thanks!
>
> Anastasia
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20181029/3dad8de1/attachment.html>